The role of a newly identified SET domain-containing protein, SETD3, in oncogenesis

Abstract

The SET domain is found in histone methyltransferases and other lysine methyltransferases. SET domain-containing proteins such as MLL1 play a critical role in leukemogenesis, while others such as SETD2 may function as a tumor suppressor in breast cancer and renal cell carcinoma. We recently discovered that SETD3, a well-conserved SET domain-containing protein, was involved in a translocation to the immunoglobulin lambda light chain locus in one of the non-homologous end-joining/p53-deficient peripheral B-cell lymphomas. We showed that a truncated mRNA lacking the SET domain sequences in Setd3 gene was highly expressed in the lymphoma. Furthermore, we found that the truncated SET-less protein displayed oncogenic potential while the full length SETD3 protein did not. Finally, SETD3 exhibits histone methyltransferases activity on nucleosomal histone 3 in a SET-domain dependent manner. We propose that this newly identified Setd3 gene may play an important role in carcinogenesis.

Introduction

Genomic DNA in the nucleus is packaged into the nucleosome, which consists of an octamer of histones (two of each H2A, H2B, H3, and H4) wrapped around with 146bp genomic DNA.1 The amino-terminal tails of the core histones can be modified by posttranslational modifications such as lysine (K) methylation (me), which is mediated by histone methyltransferases (HMTs).2 The suppressor of variegation, enhancer of zeste, trithorax (SET) domain was first recognized as a conserved feature in chromatin-associated proteins3,4 and a number of SET domain-containing proteins have since been characterized as HMTs.5,6 For example, SUV39H1, SUV39H2, and SET9 methylate H3 or H4 and function as important epigenetic regulators of mammalian development, heterochromatin, and genomic instability.7,8 Deregulation of SET-domain function has an important role in carcinogenesis.9,10 Mixed-lineage leukemia 1 (MLL1) is one of the best studied SET domain-containing proteins in human cancer, which is disrupted by chromosomal translocations resulting in the aberrant expression of chimeric proteins.11 Notably, all the fusion products contain the N-terminal 1400 amino acids of MLL1 and lack the SET domain.11 In addition, a common feature of SET domain-containing proteins is that their full length proteins such as RIZ1, MDS1-EVI1 and MMSET-II appear to act as tumor suppressors, while the corresponding SET domain-lacking proteins, RIZ2, EVI1, and MMSET-I, function as oncogenes.9,12 Truncation of the full length protein can be caused by either translocations or alternative splicing.12

V(D)J recombination is a site-specific DNA recombination process13 which breaks and rejoins DNA segments of antigen receptor loci in lymphocyte progenitors.13 DNA double strand breaks (DSBs) generated during V(D)J recombination are repaired by non-homologous end-joining (NHEJ). XRCC4 is an essential co-factor of DNA ligase IV (Lig4) and co-operates with Lig4 to catalyze the ligation step of NHEJ.14 We previously inactivated Xrcc4 specifically in peripheral B cells.15 Xrcc4 conditional knockout mice are not cancer-prone; however, once bred into a p53 deficient background, they often succumb to B-lineage lymphomas (termed CXP lymphomas) that frequently harbor clonal translocations.16 In this study, we present a SET domain-containing protein, SETD3, which is involved in a translocation to the immunoglobulin lambda (Igλ) light chain locus in one of the CXP lymphomas.

Design and Methods

Cytogenetic assays, SKY, FISH and chromosomal painting

Preparation of metaphase chromosomes, spectral karyotyping (SKY), fluorescence in situ hybridization (FISH), and whole-chromosome painting using single-chromosome-specific paints were performed as previously described.16 FISH probes were as follows: BACs for Setd3 on chr12 are RP23-360E20, RP23-42A17, and RP23-32G7 obtained from the BACPAC CHORI database.

Further details of Design and Methods are available in the Online Supplementary Appendix.

Results and Discussion

Identification of Setd3 from a NHEJ/p53-deficient peripheral B cell lymphoma

We initially observed a t(12;16) translocation by SKY in the CXP163 tumor (Online Supplementary Figure S1).16 Next, we attempted to map the junction of this translocation by FISH. Cytogenetic analysis using bacterial artificial chromosomes (BACs) that contain mouse genomic sequences covering Bcl11b or Cyclin k (Ccnk) and whole chromosomal paint showed that the translocation breakpoint of t(12;16) in CXP163 was between Bcl11b and Ccnk, and SET domain containing 3 (Setd3) is the only gene in this region (Figure 1A). In addition, these analyses revealed a complex t(16;12;16) translocation in CXP163, which involved chromosome 16 (Chr16) and appeared karyotypically normal by SKY and chromosomal painting. FISH using serial BACs covering a region on Chr12 containing Setd3 and Bcl11b, and other analyses showed that an approximately several hundred kb sequence from this portion of Chr12 was inserted into the Igλ locus on Chr16 to generate the complex t(16;12;16) translocation (Figure 1A). Thus, CXP163 contains a t(12;16) and t(16;12;16) complex translocation, both of which involve Setd3.

Figure 1.Molecular cloning of Setd3 from an Igλ translocation derived from Xrcc4/p53 peripheral B-cell lymphoma. (A) Cytogenetic analysis of t(12;16) and complex t(16;12;16) translocations in CXP163. 3 BACs covering the Setd3 region (E20, A17 and G7) and 1 BAC covering the CcnK region were used for FISH analysis. t(12;16) translocation occurred between BAC G7 and CcnK, involving the Setd3 gene. The complex t(16;12;16) translocation has a large region of Chr12 including all 3 BACs (E20, A17 and G7) covering the Setd3 gene inserted into Chr16. Chr16 was painted light green by whole chromosomal painting. (B) Southern blot analysis of CXP162 and CXP163. Clonal Igλ rearrangements were detected via Cλ1 probe. Germline (GL) bands are indicated by arrow. (C) Schematic map of t(16;12;16) translocation depicts the translocation breakpoints located in the sequence upstream of Jλ1 and in intron 8 of Setd3 gene. The inserted region of Chr12 is shown in red and Chr16 in green. Arrow indicates the transcriptional orientation of the truncated Setd3. Dotted arrow indicates the transcriptional orientation of the germline Jλ1. Question marks indicate that the translocation breakpoint at this junction remains unknown. (D) The translocation junction was cloned by phage library and sequenced to map the breakpoints in the Jλ1-Cλ1 cluster and in intron 8 of the Setd3 gene. Microhomology at the junctions is underlined. There were mutations and deletions identified at the junction. (E) Upregulation of Setd3-S in the CXP163 lymphoma by Northern blot analysis. RNA from wt splenocytes, CXP163, and CXP221 was purified and employed for Northern blot with Setd3-S cDNA as the probe. Arrows indicate the Setd3-L and Setd3-S transcripts, respectively, with the GAPDH transcript as a loading control. (F) Western blot was performed using cell extracts from 293T cells transfected with vectors expressing Flag-tagged SETD3-S, SETD3-L, or control vector with anti-Flag antibody (Ab).

Southern blotting analyses showed that the CXP163 lymphomas had Igλ rearrangements (Figure 1B). To clarify the nature of these Igλ rearrangements, we cloned them from genomic phage libraries generated from CXP163 tumor DNA. Molecular cloning revealed the translocation breakpoint that fused a portion of the Setd3 gene into the Jλ1-Cλ1 region containing the 3′Igλ enhancer (Figure 1C). The translocation junction occurred within intron 8 of Setd3 and fused Setd3 exons 9 through 13 to the intronic sequences just upstream of the Jλ1-Cλ1 region in a head-to-head configuration (Figure 1C and D). In this configuration, sense strands of these two genes are in opposite orientation, thereby excluding the possibility of forming a fusion product. Identifying the breakpoint of this complex Igλ translocation in CXP163 was part of a systematic study to clone the Igλ translocations/rearrangements in CXP lymphomas.16 These studies eventually showed that CXP lymphomas arose from peripheral B cells that had attempted secondary V(D)J recombination of Igκ and Igλ light chain genes. Correspondingly, CXP tumors frequently harbored large chromosomal deletions or translocations involving Igκ or Igλ.16 The t(12;16) translocation in the CXP163 lymphoma is not a balanced translocation, and the remainder of the distal portion of Chr12 was lost, as shown by SKY (Online Supplementary Figure S1). Notably, the genomic configuration of the region surrounding Setd3 on mouse Chr12 is highly conserved with human Chr14q32.2, a critical region frequently involved in lymphomagenesis.17 Thus, it is likely that there is a potential oncogenic consequence of the distal 12q deletion caused by the t(12;16) translocation.

Characterization of the Setd3 gene in tumor and normal tissues

The mouse Setd3 gene is normally located on the telomeric end of Chr12 containing 13 exons (Online Supplementary Figure S2A). This Igλ translocation disrupted the Setd3 gene, and the breakpoint occurred in intron 8, which is 1897bp upstream of exon 9. Using Northern blot analysis, we identified a truncated Setd3 transcript (designated as Setd3-S) which accumulated to a much higher steady state level in CXP163 than the full length Setd3 transcript (designated as Setd3-L) in wild-type (wt) splenic B cells or in other CXP tumors lacking this translocation (Figure 1E). Presumably, the putative promoter that drives the expression of Setd3-S is located within the 1897bp upstream of exon 9 of Setd3. In addition, the 3′Igλ enhancer in close vicinity might also contribute to the high expression of Setd3-S transcript. Next, we cloned the cDNA of Setd3-S using primers specific to the 5′ end of exon 9 and 3′ end of exon 13. This particular cDNA contains an open reading frame with the start codon (ATG) located within exon 10. Thus, we concluded that this Igλ translocation led to the production of a truncated Setd3 transcript encoding the C-terminal portion of Setd3.

Figure 2.Oncogenic potential of the truncated SETD3-S. (A) Left: NIH3T3 cell lines were infected with different retroviruses expressing either RAS^V12 (positive control), full length SETD3-L, truncated SETD3-S, or empty vector as a control. Mock infection control is also included. A soft agar assay was performed and colonies were counted 3–4 weeks after plating (n=4 plates per group). Representative data are shown as mean±S.D. from 3 independent experiments. (Right) Western blot for the expression of Flag-tagged SETD3-S or SETD3-L and HA-tagged RAS. (B) Representative images of colonies from all different groups are shown. Only SETD3-S or RAS^V12 infected 3T3 cells had the ability to form colonies (anchorage-independent growth). (C) Setd3-S co-operates with c-myc to increase colony formation of primary MEFs. Primary MEFs were isolated from wt embryos and infected with different retroviruses expressing either c-MYC, full length SETD3-L, truncated SETD3-S, or empty vector as a negative control individually, or simultaneously with c-MYC and SETD3-L, or c-MYC and SETD3-S. (Left) A soft agar assay was performed and colonies were counted 3–4 weeks after plating (n=4 plates per group). Data are shown as mean±S.D. from 3 independent experiments. Statistical analysis: t-test, one-tail distribution with equal variance. (Right) Western blot for the expression of Flag-tagged SETD3-S or SETD3-L and HA-tagged c-MYC.

Lack of further material precluded the characterization of SETD3 protein in CXP163. To investigate whether both Setd3 transcripts encode proteins with proper size, a Flag-tag was added to the C-terminal end of the Setd3-L and Setd3-S cDNAs. Cell extracts from Setd3-L, Setd3-S, or control vector transfected 293T cells were employed for Western blot analysis to detect the SETD3 proteins. Our data confirmed that Setd3-L and Setd3-S cDNAs gave rise to proteins with the predicted size, about 67kd and 30kd, respectively (Figure 1F). There are multiple variants of Setd3 transcripts identified in the database (see Online Supplementary Appendix) which are generated via alternative splicing.

We next examined the Setd3 transcript in various tissues by Northern blot and found that the Setd3-L transcript was present ubiquitously in all the tissues tested using the GAPDH transcript as control (Online Supplementary Figure S2B and C). The Setd3 gene encodes a protein with a highly conserved SET domain located at the N-terminus and a well-conserved RuBisCo lysine methyltransferase (LSMT) C-terminal substrate-binding domain (Online Supplementary Figure S3). Interestingly, the truncated SETD3-S protein lacks the highly conserved SET domain. The human homolog of SETD3 protein is 92% identical to its mouse counterpart (NCBI_Blastp). Thus, the Igλ translocation revealed a novel SET domain-containing protein, SETD3, which was also recently identified by independent studies using a bioinformatics approach.18,19

Role of Setd3 in cell growth

Since the expression of the truncated Setd3-S transcript was much greater in the CXP163 tumor, we next tested whether SETD3-S has oncogenic potential using a soft agar assay. Normally, NIH3T3 cells can not undergo anchorage independent growth and do not form colonies in soft agar. To test whether SETD3-S or SETD3-L can induce colony formation of NIH3T3 cells, we infected these cells with retroviruses expressing either Setd3-S or Setd3-L cDNA with a Flag-tag at the C-terminal end, or empty vector as a negative control. These analyses revealed that the cells infected with Setd3-S can form colonies at a much higher level compared to cells infected with empty vector or uninfected cells (Figure 2A and B). In contrast, the colony number in the Setd3-L infected group was similar to that in control groups (Figure 2A and B). The cells infected with retrovirus expressing Ras with a C-terminal HA-tag served as a positive control that induced robust colony formation of 3T3 cells (Figure 2A and B). Western blot analysis showed the expression of corresponding exogenous proteins in the different groups of infected cells (Figure 2A). Thus, we conclude that the truncated SETD3-S has an oncogenic potential to transform NIH3T3 cells, whereas the full length SETD3-L does not. Our data also suggest that the highly expressed Setd3-S in CXP163 might contribute to lymphomagenesis. In this context, the CXP163 lymphoma not only highly expresses truncated Setd3-S transcript (Figure 1E) but also dramatically up-regulates expression of the c-myc oncogene16 which is due to a t(12;15) translocation (Online Supplementary Figure S1) juxtaposing the c-myc oncogene next to the 3′Igh regulatory region.16 Thus, we reasoned that the truncated Setd3-S, possibly by disrupting Setd3-L function, might co-operate with c-myc to promote oncogenesis. Next, we tested whether co-expression of c-myc and Setd3-S in primary mouse embryonic fibroblasts (MEFs) promotes colony formation. Primary MEFs were infected with retroviruses expressing Setd3-S, Setd3-L as described above, or c-myc with a C-terminal HA-tag, or empty vector. Western blot analysis showed the expression of exogenous proteins in the corresponding groups of infected 3T3 cells (Figure 2C). Our data showed that c-myc expression had a moderate effect on colony formation of primary MEFs, whereas both Setd3-S and Setd3-L had no obvious effects compared to the control group (Figure 2C). In contrast, when c-myc and Setd3-S were co-expressed, the number of colonies was significantly increased, whereas co-expression of Setd3-L with c-myc had no such effects (Figure 2C). Thus, our data demonstrate that Setd3-S may have the potential to function as an oncogene.

Figure 3.In vitro HMT assay for SETD3. (A) Coomassie blue staining of the SDS-PAGE gel shows the purified GST-SETD3-L, GST-SETD3-S, and GST only proteins, with the protein marker loaded on the left (M). (B) In vitro HMT assays for GST-SETD3-L, GST-SETD3-S, and GST. Nucleosomes purified from HeLa cells were used as the substrate. Individual recombinant protein was added into the HMT assay as indicated (+). The final reaction was analyzed by SDS-PAGE followed by autoradiography. Methylation of H3 is indicated with an arrow.

Functions of SETD3 in histone modification

To investigate whether SETD3 has any SET domain-dependent enzymatic activity on histones, we performed in vitro HMT activity assays. First, we expressed and purified recombinant Glutathione S-transferase (GST) only, GST-SETD3-L, and GST-SETD3-S proteins (Figure 3A). These purified proteins were assayed for their putative methyltransferase activity using native HeLa nucleosomes, containing all histones except H1, as substrates.20 We found that only the purified GST-SETD3-L protein showed robust H3-specific HMT activities on the substrates (Figure 3B). In contrast, GST-SETD3-S or GST proteins displayed no detectable level of HMT activity (Figure 3B). When free histone octamers were used as substrates, we could not detect any HMT activity (data not shown), suggesting that SETD3-L only functions on the physiologically relevant substrates, i.e. native nucleosomes, of which H3 is a component. Thus, we conclude that the SETD3 protein is a fully active HMT for H3 with its activity dependent on the SET domain. Recently, independent studies also identified the zebrafish and mouse Setd3 genes via a bioinformatics approach18,19 which also showed that SETD3 has HMT activity, in particular, on H3K4 and H3K36.18 These studies used several H3 peptides instead of nucleosomes as substrates for the HMT assay. However, we could not detect HMT activity when free histone octamers were employed as substrates. It is likely that the activity of SETD3 on native nucleosomes, which are physiological substrates in vivo, is higher than that on free histone substrates. It remains to be determined whether SETD3 displays the same substrate specificity on native nucleosomes. In addition, SETD3 contains another well-conserved RuBisCo LSMT C-terminal substrate-binding domain. Rubisco LSMT is a chloroplast-localized SET domain-containing protein, which catalyzes the trimethylation of K14 in the large subunit of Rubisco,21 an essential photosynthetic enzyme.22 The RuBisCo LSMT C-terminal substrate-binding domain is also present in ribosomal lysine (K) methyltransferase (RKM) 1–4 of Saccharomyces cerevisiae.23 RKM1 has been shown to methylate the ribosomal 23 subunit.24 Thus, it remains possible that SETD3 might also have other non-histone substrates in the cytoplasm. Taken together, we propose that SETD3-L might function through its well-conserved SET domain and play an important role in suppressing tumor development while SETD3-S might function as a dominant negative mutant of SETD3-L and promote oncogenesis. Thus, it would be of interest to investigate the potential involvement of the Setd3 gene in carcinogenesis.

Acknowledgments

We thank Dr. FW Alt (Children’s Hospital Boston, HHMI, USA) for his generous support of this project and Dr. Y. Geng (Dana-Farber Cancer Institute, Boston, MA, USA) for advice on the soft agar assay and vectors including pBabe-puro-Ras and pBabe-hygro-c-myc. We thank M Murphy for technical support with FISH and SKY. We apologize to those whose work was not cited due to length restrictions.

Footnotes

• The online version of this article has a Supplementary Appendix.
• Funding This work was supported by University of Colorado School of Medicine startup funds and Leukemia Research Foundation funds (to JHW). JHW is a recipient of the Boettcher Foundation Webb-Waring Biomedical Research Award. CTY is a recipient of the V Foundation Scholar Award, Emerald Foundation New Investigator Award, Kimmel Foundation Scholar Award, and ASH Scholar Award.
• Authorship and Disclosures Information on authorship, contributions, and financial & other disclosures was provided by the authors and is available with the online version of this article at www.haematologica.org.

• Received March 28, 2012.
• Accepted September 26, 2012.